Find orphan pages your sitemap does not explain

Compare what should exist with what readers can reach

A page can appear in a sitemap while no normal site page links to it. It may be discoverable to a crawler yet remain difficult for a reader to find. Start with two inventories: intended public canonical URLs and the internal links observed from accessible pages.

Google recommends crawlable links using ordinary anchors with href destinations and useful link text. A link graph should inspect that actual navigation evidence, not only a list of URLs returned by the CMS. [Google Search Central: Link best practices]

A reproducible small graph audit

1. Export intended public URLs with canonical, purpose and indexability policy.
2. Crawl the authorized public scope without forms, private APIs or account actions.
3. Extract internal href targets and normalize redirects/canonical families carefully.
4. Record source URL, destination, anchor and link location.
5. Compare each intended page with its observed incoming-link evidence.
6. Review important sitemap-only or disconnected pages and the crawler’s coverage limits.

Separate contextual body links, hub cards, breadcrumbs and sitewide navigation. A footer link can make a page reachable without explaining why it helps with a particular task. Conversely, one well-placed contextual link can be more useful than many repeated boilerplate links.

Inventory: /insights, /tools, /tools/cost, /guides/cost, /guides/refunds
Observed links:
  /insights → /guides/cost
  /guides/cost → /tools/cost
  /tools → /tools/cost
Candidate for review: /guides/refunds has no observed incoming edge

The fixture identifies a candidate within the observed graph. It does not prove that no link exists anywhere. A crawler can miss a page because of scope, access, rendering, extraction limits or a failed request. Save those limitations instead of declaring every unseen edge absent.

Add the path that fits the task

For the synthetic refund guide, a relevant revenue-calculation article or resource hub could explain when the reader needs refund reconciliation and link to it. A random link from every unrelated post would make the graph denser while weakening editorial focus.

Check the destination before linking. A useful incoming edge to a noindex draft, a missing resource or an unrelated redirect does not satisfy the intended discovery policy. Align canonical addresses so that navigation does not fragment between equivalent URL forms.

• Use descriptive anchors that explain the next task.
• Link from the closest existing article or hub with a genuine relationship.
• Keep the destination useful as a direct entry from search.
• Verify reciprocal navigation where it helps readers, without forcing every page to link back to everything.

Recheck the graph and the real journey

After the change, request the source page and follow the new link on desktop and mobile. Verify the target status, canonical and visible answer. Then rerun the bounded graph comparison to confirm the expected incoming edge is present.

I would report the count and purpose of repaired important pages alongside the crawl scope. A sitemap accepted by a search engine and a live contextual navigation path are separate outcomes. Neither guarantees indexing or traffic, but both can be checked honestly.

• Review the content-hub model
• Organize related search tasks
• Review a proposed destination

Sources

AI-assisted writing and an AI-generated editorial cover. Source links and checked dates are listed below; illustrative examples are labeled in the text.

• Google Search Central: Link best practices — checked 2026-10-18